2stageEF high performance class AB power amp / 200W8R / 400W4R

Jump to Latest
astx

astx

Member
Joined 2011
Paid Member
keantoken said:
Try the Sanken 2SC4883A/A1859A which can still be gotten from Digikey. My models cover quasi-saturation:
...
Click to expand...

THX for the models! A little bit higher Cob and lower Vce but maybe a good (and expensive) future replacement for the discontinued 2SC4793/2SA1837.
Due to the lower Vce maybe only usable for SA2015.
Funny: on the Digikey shop there is no information about discontinuation of 2SC4793 - only for the PNP part 2SA1837 with a watermark in the datasheet ...

Have Fun, Toni
 
astx

astx

Member
Joined 2011
Paid Member
keantoken said:
I was warned away from Unisonic transistors when they were being sold at Profusion by homemodder who said they were slow and inferior. This was for the D669/B649 pair but I imagine it applies to many of the others.

But, it would be nice to see some actual comparisons.
Click to expand...
Thanks for info/warning. I've ordered 10 pieces
"2SC4793LB-TF3T, 2SA1837LB-TF3T" each from profusion for testing.

BR, Toni
 
astx

astx

Member
Joined 2011
Paid Member
keantoken said:
Ian Hegglun's cube-law amp uses 470R gate stoppers for the 20X20.

The amount needed probably depends a lot on what is driving the FETs, so you might be fine with more or less.
Click to expand...

IMHO: a rather high value. I would have expected lower values due to lower fT compared to vertical mosfets but:


  • AN1850/LME49830/snaa058a.pdf recommends 22R for 2SK1530, 10R for 2SJ201. (Fig.3, page 3)
  • Bob Cordell uses 470R with 2SK1056/2SJ160 (laterals) (DAPA, fig. 11.1/p.216)
  • Bob Cordell uses e.g. 100R for verticals 2SK1530, 2SJ201 (DAPA, fig. 11.17, p.242). He also recommends to add a gate/drain zobel (e.g. 100R + 47pF) instead of reducing the speed of the mosfets
and an interesting application note for laterals from semelab:
http://products.semelab-tt.com/pdf/ApplicationNoteAlfet.pdf
p.11: "Gate Resistor Recommendations": 150R up to 470R.

Maybe a combination of a lower gate resistors and "cordell" gate zobels is the best way?

BR, Toni
 
keantoken

keantoken

Member
Joined 2006
astx said:
THX for the models! A little bit higher Cob and lower Vce but maybe a good (and expensive) future replacement for the discontinued 2SC4793/2SA1837.
Due to the lower Vce maybe only usable for SA2015.
Funny: on the Digikey shop there is no information about discontinuation of 2SC4793 - only for the PNP part 2SA1837 with a watermark in the datasheet ...

Have Fun, Toni
Click to expand...

Those models have been updated with correct Vaf values. The test didn't have very good resolution at high Vaf, so they could be a bit higher than that. In the past I may have sent you previous models with incorrect Vaf.
 
astx

astx

Member
Joined 2011
Paid Member
keantoken said:
Those models have been updated with correct Vaf values. The test didn't have very good resolution at high Vaf, so they could be a bit higher than that. In the past I may have sent you previous models with incorrect Vaf.
Click to expand...

Sorry - it was not my intention to criticize your bjt models. I meant the difference between the datasheets ... ;)
I will check my "keantoken_bjt_models_standard.lib" and add the new models. Thank you again for your excellent work providing us updated and new spice models! :)

Br, Toni
 
keantoken

keantoken

Member
Joined 2006
I do encourage you to find out on the bench what are acceptable gate stopper values, because often it's not the FET itself that becomes unstable, but the amplifier driving it. The stoppers can often be lowered quite a bit with good layout and optimized driving circuitry, or with a well-placed RC snubber.

I updated the Cjc of the P Sanken driver. The new models are in my latest model library, attached. There are several improvements:

Code: 
* Version 9:
*
* 2SC4883A/A1859A: 	Updated Vaf and Cjc with measured values.
* 2SC5171/A1930:	Updated with quasi-saturation.
* 2SC3423/A1360:	Updated with more reasonable Vaf values, but need measurement.
* BC3x7-40:		Updated with measured Vaf values.
* KSC1845E:		Updated with measured Vaf values and quasi-saturation.
* KSA992F:		Updated with quasi-saturation.
* BC5x0C:		Updated with measured Vaf values for different factories.
* MPSA18/MMBT6429	New model with measured Vaf for a Fairchild MPSA18 and Rb according to the OnSemi datasheet.
 

Attachments

  • kean9.bjt.txt
    55.4 KB · Views: 141
astx

astx

Member
Joined 2011
Paid Member
keantoken said:
I do encourage you to find out on the bench what are acceptable gate stopper values,
...
Click to expand...
Will do so. The annotation note of semelab presents a formula to pre-calculate such a resistor. The rest is practical work. I own a 150MHz scope so I should be able to check the oscillation sensitivity...

keantoken said:
I updated the Cjc of the P Sanken driver. The new models are in my latest model library, attached. There are several improvements:
...
Click to expand...
Many thanks - just downloaded!
BTW: the BC560 models still have hFe(BF?) of 900 which I have never seen. Most of my 2k Onsemi BC560CG are in the 600 range and most of my newest 2k Fairchild BC560CTA are in the 450 range.

stewin said:
nice buid astx
Click to expand...
Thx!

Have fun, Toni
 
keantoken

keantoken

Member
Joined 2006
Do the BC560 show too high Hfe in simulation? Sometimes Bf can be higher than the actual beta of the transistor, in order to model an upward slope at low currents better.

Those models are based on Cordell's models, which may have been made for atypical devices.
 
Last edited:
astx

astx

Member
Joined 2011
Paid Member
keantoken said:
Do the BC560 show too high Hfe in simulation? Sometimes Bf can be higher than the actual beta of the transistor, in order to model an upward slope at low currents better.

Those models are based on Cordell's models, which were made for atypical devices it seems.
Click to expand...
Hmmm ... OK ... overflow ... sorry ... lack of knowledge here on this side ... thought spice parameter BF = hFe
:):crazy:
 
A

AndrewT

R.I.P.
Joined 2004
keantoken said:
Do the BC560 show too high Hfe in simulation?
Click to expand...
Yes
Sometimes Bf can be higher than the actual beta of the transistor, in order to model an upward slope at low currents better.

Those models are based on Cordell's models, which may have been made for atypical devices.
Click to expand...
And yes, they don't come out quite as high as Bf would indicate, but still far lower base current is predicted compared to the 550/560C that I use.
 
astx

astx

Member
Joined 2011
Paid Member
SA2013, SA2014, SA2015 housekeeping circuits

SA2013, SA2014, SA2015 housekeeping circuits

I've been asked several times mostly via PM to provide more information about the housekeeping circuits which I've developed and used in most of my current amplifier projects. If there's enough positive response here, I will think about providing gerbers and maybe firmware for programming the pic16f690 and the newer replacement pic16f1829...

Have fun, Toni
 
astx

astx

Member
Joined 2011
Paid Member

Attachments

  • utc_4793.png
    utc_4793.png
    45.4 KB · Views: 554
  • tho_4793.png
    tho_4793.png
    45.5 KB · Views: 554
  • utc_1837.png
    utc_1837.png
    43 KB · Views: 562
  • tho_1837.png
    tho_1837.png
    46.9 KB · Views: 555
  • measurements.png
    measurements.png
    27.3 KB · Views: 548
You must log in or register to reply here.
Top Bottom